Diffei Academic | mNGS effectively identifies pleural effusion: assists in the diagnosis of tuberculous pleurisy and malignant tumors

Introduction

Recently, the Pathology Department of Beijing Chest Hospital Affiliated to Capital Medical University collaborated with Diffei Medical to publish an article in the Journal of Translational Medicine (IF=7.4).This study compared the performance of mNGS and conventional tuberculosis detection methods in 138 patients with pleural effusion. It also made full use of the human sequences and microbial sequences generated by mNGS to explore the value of this technology in infectious diseases and the identification of malignant tumors.

Research Background

Pleural effusion (PE) is a common complication of various diseases, and the diagnosis of its cause is challenging.Common causes include Mycobacterium tuberculosis, parapneumonic pleural effusion, and malignancy.

1. The positive rate of acid-fast staining of tuberculous PE is very low, and the positive rate of Mycobacterium tuberculosis culture in PE is 40%.Molecular diagnosis has high specificity but low sensitivity.

2. Parapneumonic PE is often secondary to bacterial pneumonia, lung abscess or bronchiectasis combined with infection.Positive PE microbiological culture results are the gold standard for diagnosing parapneumonic PE, but the culture cycle is long and the positive rate is about 37.7% to 53.5%.

3. The sensitivity of PE cytology in diagnosing malignant tumors is 40.0% to 87.0%. In addition, PE can also be used as a liquid biopsy sample for metagenomic next-generation sequencing (mNGS) testing to detect genomic instability. sex, an important genetic characteristic of malignant tumors, for diagnosis.

Therefore, the use of mNGS has become a useful tool to identify common causes of pleural effusion.

Research Highlights

1. Use 5 detection methods to diagnose tuberculous PE.mNGS was found to exhibit excellent sensitivity of 49% and specificity of 100%.

2. Using human sequences for copy number variant (CNV) analysis to diagnose malignant PE (MPE), the sensitivity was 54.1%, the specificity was 80.8%, the accuracy was 71.7%, and the AUC was 0.851.

research route

A total of 138 patients were included in the study, including 82 patients with tuberculous pleurisy (TBP), 35 patients with MPE, and 21 patients with non-tuberculous infection.Use the microbial sequences generated by mNGS to evaluate the diagnostic value of mNGS for tuberculosis.At the same time, the human sequence generated by mNGS was used to evaluate the diagnostic value of mNGS for MPE (Figure 1).

Figure 1 Workflow and enrollment queue display of mNGS for tuberculosis detection and malignancy prediction on PE samples

Research result

Performance comparison of different tuberculosis diagnostic methods, mNGS has excellent performance

This study used a variety of clinical detection methods, including culture, AFB test, Xpert and T-SPOT, and mNGS for tuberculosis detection.Among all clinical tests, the T-SPOT test showed the highest positive detection rate, 61% (culture 32%, Xpert 35%, AFB 12%), which was also slightly higher than mNGS (49%, Figure 2).However, the false positive rate of the T-SPOT method is as high as 11%, which is much higher than other detection methods.In contrast, mNGS demonstrated excellent sensitivity (49%) and specificity (100%).

Figure 2 Tuberculosis detection results between mNGS and conventional clinical experiments

Using mNGS human sequences to construct a CNV model to identify MPE

To evaluate the value of mNGS in identifying malignant tumors, this study used the GLM method to construct an mNGS-CNV model to identify MPE.Twenty-five cases of non-malignant pleural effusion (non-malignant PE, non-MPE) were randomly selected as the baseline, and the chromosome copy number of the remaining 113 patients (35 MPE and 78 non-MPE) patients was analyzed and a copy number variation map was drawn. , using the constructed mNGS-CNV model to identify MPE and non-MPE (Figure 3A).Compared with clinical pathological diagnosis, the sensitivity of the mNGS-CNV model was 51.4%, the specificity was 80.8%, and the accuracy was 71.7% (Figure 3B, C).

Subsequently, the prediction performance of the mNGS-CNV process model was evaluated through LOOCV, and the AUC value of the ROC curve was 0.851 (the closer the AUC value is to 1, the better the model performance; conversely, the closer the AUC value is to 0, the worse the model performance is, Figure 3D ), showing the excellent performance of this prediction model.

Figure 3 Diagnostic performance of mNGS-CNV analysis.A. Representative chromosomal copy number variation diagram of MPE patients; B. Comparison of mNGS-CNV modeling results and clinical pathological diagnosis; C. Sensitivity, specificity and accuracy of mNGS-CNV prediction based on diagram B; D . Performance of mNGS-CNV LOOCV results.

Summarize

This study made full use of the microbial sequences and human sequences generated from mNGS. The analysis results showed that mNG has good identification ability of TBP and MPE, which provides the possibility for mNGS to perform pathogen detection and tumor diagnosis at the same time. However, prospective clinical trials are still needed in the future. Research and large-sample validation.